Joy is Not Made to Be Prompt
Why the human perspective matters in the age of LLMs

Within the academy, a phrase we come across often these days is “AI-generated”. To describe Large language models (LLMs), in training sessions and otherwise, there is another phrase that comes up, “Generative AI”. Drawing parallels to human thinking, our reading and education builds our capacity to generate new thoughts, ideas and meanings. We can be drawn into something generatively – opening up our minds to possibilities and novel approaches, including in prompting LLMs. But, for something to strike as generative, connection is central to the experience.
I came across a poem, titled, “Conditions of Travel”. The title conjurers an image in my mind. I read further, and the poem expands my image.
The poet, Katherine Stansfield, associates travel a bus, whereas I associated travel with airplanes:
‘That when I lose track of time and fear the last bus already been and gone, my walk home in the dark a trap waiting to spring, a bus will come, its destination a proclamation of light, its destination the place I need: Nan’s house, the rainforest, the day in 1994 on the pier when he said Yes over the best chips ever fried.’
Reading the poem draws out worlds, old and new. The lines are so specific to her experience and location in Britain, and yet I can relate it to my daily travel, while making new meanings: The bus brings up my childhood visits to Mumbai, and the long trip from Lokhandwala to Colaba on a double decker, visiting relatives and friends. It illuminates, for me, how much has changed with time – now I take an uber. It pushes me to think about the joy (and pain) of mundanity, and what that experience may be for others – with our cities evolving into large, cosmopolitan beings, with burgeoning informality in a time of climate change and economic instability.
I experiment. I prompt a chat-bot to write me a poem on the topic, and it generates the following lines:
‘The road is never as it seems:
It bends through traffic, dust, and dreams.
A delayed train, a sudden rain,
a window blurred by joy and pain.‘
I discuss these lines alongside Stansfield’s poem in a workshop. A participant shares, it sounds nice, but empty.
The AI-generated poem is true to the purpose of LLMs – efficiency and task-orientation. But it collapses travel into widespread cliches, universally applicable. It collapses joy to a prompt, a crumb – paraphrasing Mary Oliver – Joy is not made to be a crumb.
Joy, like creativity and critical thinking, is made of details, of stories, of connections often emerging unexpectedly. In working with LLMs, we generate with a focus on certainty and predictability.
The future of our world, however, is uncertain and unpredictable, ever-changing with complex crises besetting our planet. Experiences of these challenges are varied and distinct. Our perspective, individual and collective,enlivens details that can otherwise be erased, compromising what can generate and deepen interdisciplinarity and creativity in addressing challenges.
In the age of LLMs, the human perspective matters precisely because it our perspective that enables us to rewrite the stories we have inherited and rebuild with other, novel possibilities. After all, it is our perspective that creates the prompts we work with in LLM’s, prompts we must draw upon to think further, deeper.
Our perspective matters because joy and our experience of the world, is more than a prompt.
~ Vrinda Chopra, Senior Writing Fellow, Ashoka University